digitally tunable physicochemical coding of material ...tissueeng.net/lab/papers/nmat3108.pdf ·...

LETTERSPUBLISHED ONLINE: 4 SEPTEMBER 2011 | DOI: 10.1038/NMAT3108

Digitally tunable physicochemical coding ofmaterial composition and topography incontinuous microfibresEdward Kang1, Gi Seok Jeong1, Yoon Young Choi1, Kwang Ho Lee1,2†, Ali Khademhosseini3,4,5

and Sang-Hoon Lee1*

Heterotypic functional materials with compositional and topo-graphical properties that vary spatiotemporally on the micro-or nanoscale are common in nature. However, fabricating suchcomplex materials in the laboratory remains challenging. Herewe describe a method to continuously create microfibres withtunable morphological, structural and chemical features usinga microfluidic system consisting of a digital, programmableflow control that mimics the silk-spinning process of spiders.With this method we fabricated hydrogel microfibres codedwith varying chemical composition and topography along thefibre, including gas micro-bubbles as well as nanoporousspindle-knots and joints that enabled directional water col-lection. We also explored the potential use of the coded mi-crofibres for tissue engineering applications by creating mul-tifunctional microfibres with a spatially controlled co-cultureof encapsulated cells.

Micro- and nanoscale fibres have attracted substantial attentionowing to their extensive applications1–4. Several methods haverecently been developed to fabricate small fibres continuously, suchas electrospinning andmicrofluidic spinning5–8. Most continuouslygenerated fibres have been cylindrical in shape with a homogeneouschemical composition, and the creation of continuous microfibreswith spatiotemporal variations in chemical composition and mor-phology has been a great challenge. The fibre-spinning mechanismof spiders may provide critical clues to addressing this challenge.The main features of this spinning process include its small scale,simplicity and tunability of the produced fibres. Microfluidictechnology is one potential candidate that can achieve all of thesefeatures. Althoughmicrofluidic chips have been used extensively forfabricating particles with controlled shapes9–14, the continuous pro-duction of fibres has presented a considerable challenge. Previouswork by our group and others has demonstrated the feasibility ofusing microfluidics to engineer continuous fibres8,15,16. Recently,we reported the simultaneous production of multiple fibres,each of differing composition, using coaxial polydimethylsiloxane(PDMS) microfluidic channels17. Although these systems havebroken several barriers of conventional methods, it has not beenpossible to generate fibres with complex morphologies and tunablecomposition to the same extent as those present in nature.

1Department of Biomedical Engineering, College of Health Science, Korea University, Jeongneung-dong, Seongbuk-gu, Seoul 136-703, Republic of Korea,2Department of Mechanical Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, Massachusetts 02139, USA,3Center for Biomedical Engineering, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, 65 Landsdowne Street,Cambridge, Massachusetts 02139, USA, 4Harvard-MIT Division of Health Sciences and Technology, Massachusetts Institute of Technology, Cambridge,Massachusetts 02139, USA, 5Wyss Institute for Biologically Inspired Engineering, Harvard University, Boston, Massachusetts 02115, USA. †Presentaddress: Department of Advanced Materials Science and Engineering, Kangwon National University, Hyoja 2-dong, Chuncheon-si, Gangwon-do, 200-701,Republic of Korea. *e-mail: [email protected].

Here, we propose a spinning method that uses a microfluidicchip combined with a digital fluid controller. By introducing activefunctional modulations, such as by switching the flow throughthe use of a digital control scheme, we produced microfibresalongwhich diverse chemical compositions, structures, gas bubbles,live cells and morphologies could be spatiotemporally coded.The coding resolution across the fibre was a few micrometresand could be regulated along the fibre to a few hundredmicrometres. To our knowledge, such fibres have not previouslybeen fabricated. Furthermore, an application of these fibresfor tissue engineering was demonstrated by fabricating cell-laden microfibres with spatially regulated patterned cellular co-cultures. Similar to nature, in which functional materials aregenerated by combining diverse compositions of existing matterspatiotemporally, the proposed process may enable the creation ofnew materials by physicochemical modulation of existing materialson the microscale.

Insight from the spinning apparatus of spiders may be usedto create a biomimetic approach to generate fibres. A spiderproduces fibres by mixing fluids from glands in its abdomenand spinning the resulting mixture (Fig. 1a). At the end of itsspinning duct, a valve acts as a clamp to grip the thread oras a ratchet to restart spinning after internal rupture18. Fluidsfrom different glands lead to the spinneret so that silk blendswith specific properties, required for a particular function, can beproduced19. We used a microfluidic process comprising a numberof individually controllable inlets to generate a spinning processthat mimics the spinning process of spiders. Figure 1b,c showsa schematic and an image of a microfluidic chip designed toproduce microfibres with spatially organized morphologies andcompositions. On this fluidic platform, the sample fluid volumeof each channel is tightly controlled by a computer-controlledpneumatic valve (see Methods and Supplementary Fig. S1 fordetails). We used a thin PDMSmembrane (≈10 µm) and through areplication process we constructed hybrid channels (a combinationof rectangular and cylindrical channels) on a single device17.Integrating small pneumatic valves into the hybrid channels wasa critical technical challenge, which was addressed by using theparabolic membrane naturally generated by the surface tension

NATURE MATERIALS | VOL 10 | NOVEMBER 2011 | www.nature.com/naturematerials 877

© 2011 Macmillan Publishers Limited. All rights reserved

http://www.nature.com/doifinder/10.1038/nmat3108

mailto:[email protected]

http://www.nature.com/naturematerials

LETTERS NATURE MATERIALS DOI: 10.1038/NMAT3108

e

B′BB′B

d Parabolic membrane

Open

GasClosed

Gas

Twist

Spool

A′A Sample

A′A

a

Water, surfactant and phosphate

Protein

Spider’s valve

Spigot

Spinning duct

Glands

b cSheath inlet

Sample inlet

Valve

Air channel

Coded fibre

Rotating spool

Spinning duct, lubricant and Ca2+

Motor-driven rotation

Figure 1 | Concept of coded microfibre production. a, Anatomy of the silk-spinning system of spiders. The silk gland produces several proteins, thespinning duct focuses and solidifies the protein solution and the valve controls the flow of the protein. b, Conceptual description of the process ofgenerating coded fibres. The extruded fibres were continuously wound on the spool by a motorized system (Supplementary Movie S2). c, Photograph ofthe microfluidic spinning chip. d, Optical image of the ‘open’ and ‘closed’ states of the valve above the channel (left) and a cross-sectional schematic of thevalve operation (right). e, Schematic of the process of twisting fibres using a motor system (top) and a fluorescence micrograph of a twisted fibre with red-,green- and blue-stained fibres (bottom, Supplementary Movie S6). Scale bars, 5 mm (c), 2 mm (d) and 500 µm (e).

of the pre-polymer PDMS with air (Fig. 1d; fabrication processand valve analysis is shown in Supplementary Figs S2 and S3).The end of the chip consists of cylindrical channels with variousdiameters to generate the coaxial flow (Supplementary Fig. S4).The sheath flow (CaCl2) separates the alginate stream from thechannel wall, shapes the focused sample flow (alginic acid) andinduces subsequent sample gelation by chelation of the Ca2+ions to form a solid alginate fibre20. The sheath fluid not onlyfocuses the alginate flow, but also provides a lubricant layerduring alginate gelation (Supplementary Information). In ourexperiments, alginate was used because it has been approvedby the Food and Drug Administration for certain biomedicalapplications, although a range of other materials (for example,polyethylene glycol and chitosan) can also be used8,21,22. To controlthe alignment of the extruded fibres, they were wound arounda small rotating spool. The relationship between the flow ratesand the maximum winding velocity of the motor is shown inSupplementary Fig. S4d (see Methods for detail). Furthermore, bytwisting the fibres with a spool, triple-helix fibres were generated,as shown in Fig. 1e.

The proposed platform was used to code a range of chemicalcompositions within the fibres. To code chemical compositionsin a microfibre, the fluid volume of each inlet channel wasindependently controlled by a pneumatic valve. Figure 2a showsa control scheme for valve operation to fabricate microfibres thatcontained different materials coded into them either in a serial,parallel or mixed (that is, containing both serial and parallel

coated regions) manner. Figure 2b shows a photograph of afibre produced with both serial and parallel coding, and Fig. 2cshows a fluorescence micrograph of a serially coded fibre. Ascan be seen, the controlled opening of upstream valves yieldeda microfibre coded with different coloured dyes along its length.With the present set-up the smallest length of the serial codingwas ∼800 µm, although this is expected to improve with moreadvanced valves (Fig. 2c, inset, and Supplementary Fig. S7). Suchcoding has many potential applications for biosensing, high-throughput screening and the spatiotemporal cell positioning ofheterogeneous cells for tissue engineering. Recently, Janus fibrescoded in parallel have been reported14,23. However, serial or mixedcoding of materials using a microfluidic device has not beendemonstrated previously. Figure 2d illustrates a fibre coded withdifferent compositions in parallel across the fibre, indicating thatfunctional microscale fibres in which diverse chemicals and cellsare spatiotemporally coded may be readily produced. Although afibre with three parallel regions across its diameter is illustrated,the number and dimensions of the parallel coded regions are highlycontrollable and can be easily tuned based on the complexity of theupstream fluidic design.

We also investigated the use of spatially controlled microfibresfor recreating functional structures associated with biologicalsystems. Recently, it has been reported that a spider’s web capturesdew through microstructures on the silk fibres24. In this process,tiny water droplets initially condense on the puffs on the silk fibres,causing the puffs to shrink as water condenses to form periodic

878 NATURE MATERIALS | VOL 10 | NOVEMBER 2011 | www.nature.com/naturematerials




NATURE MATERIALS DOI: 10.1038/NMAT3108 LETTERSa

d

R GB

c

R

B

G

b

R

B

G

R

B

G

Serial coding Parallel coding

Time

Mixed coding

Open Closed

Parallel coding

Cross-section800 µm

Figure 2 | Fabrication of spatially controlled microfibres with diverse chemical compositions. a, Schematic of the digital control steps for the microfluidicspinning chip for serial, parallel and mixed coding with various composite solutions (R, alginate solution with red fluorescent polystyrene beads; G, green;B, blue). b, Mixed coding (serial and parallel) of a fibre. c, Serially coded fibre (inset, serially coded fibre on the microscale) (Supplementary Movies S7 andS9). d, Parallel coding of fibres. Scale bars, 200 µm (d), 400 µm (c, inset) and 1 mm (b,c).

spindle-knots. The linking of spindle-knots and joints, as foundin wet spider silk, contributes to directional water collection asa result of a surface-energy gradient and a difference in Laplacepressure. One of the advantages of the spinning chip developedhere is that it can generate fibres with controllable topographyand structure. The fibre diameter and length of spindle-knots canbe modulated by the valve operation (Supplementary Fig. S5).Furthermore, the porosity of the resulting fibre can be controlledby changing the material that is coded along the fibre. As shownin Fig. 3a, we introduced two alternating fluid streams to generatecontinuous fibres with evenly spaced nanoporous spindle-knots. Toconstruct a porous spindle-knot, we injected an alginate solutioncontaining salt (0.5%w/w) at a high flow rate into one of the samplechannels. It is noteworthy that the length of the spindle-knot couldbe controlled by altering the time that the valvewas kept open.

We generated nanoporous structures in the spindle-knot bysalt dissolution in the fibres (Fig. 3b). As illustrated in Fig. 3c,after leaching of salt, nanoporous structures were generated thathad a higher surface energy than joint regions, which attractedwater. At the same time, the tapered shape of the spindle-knotgenerated a difference in Laplace pressure, which provided a drivingforce for water movement. Figure 3d shows a photograph of anartificial fibre consisting of nanoporous spindle-knots and jointsthat closely resembles spider silk. To investigate the ability of theengineered fibres to mimic the water-collecting behaviour of spidersilk, drops of water were added on the fibre. As expected, thewater droplets self-collected at the spindle-knots, forming largedroplets (Fig. 3e). Interestingly, the water-collecting capabilities ofthe fibres depended on the length of the spindle-knot. Figure 3f

plots the volume of water droplets at the spindle-knot as a functionof the length of the spindle-knot. The water droplet volumecorrelated linearly to the length of the spindle-knot (r2 = 0.97).Such microfluidic coding of a heterogeneous structure may beuseful for topics ranging from the basic understanding of thephysiology of fibre spinning mechanisms in various organisms topractical applications involvingwater collection and purification.

In nature, tubuliform silk fibres, which are used to encapsulateeggs, have small grooves on their surfaces25. We fabricated artificialtubuliform fibres codedwith grooves on the surface using a groovedround channel (Fig. 3g and Supplementary Fig. S6). As can be seenin Fig. 3h, grooved fibres could be fabricated in a highly uniformmanner. Furthermore, the number and size of grooves can be tunedby changing the shape of the channel and flow rate. To assess theuse of microgrooved fibres in tissue engineering applications, ratembryonic neurons (from embryos of pregnant Sprague–Dawleyrats at 16 days of gestation) were placed on smooth and groovedfibres and analysed for their resulting cell alignment after 5 days.As shown in Fig. 3i, neurons extended neurites along the lengthof both smooth and grooved fibres; however, these extensionswere significantly more aligned and controlled along the ridges ofthe nanogrooved fibres (see Methods for details). Such controlledalignment is of great importance in directing the extension of axonsand controlling their subsequent connection with other cells, andhas important implications in tissue regeneration after spinal cordor peripheral nerve injury.

More complex coding of morphology and composition in thefibre was also achieved. A slight modification of the material andflow control scheme enabled diverse morphologies and varying






b

c

a

f

e

ig

g

Alginate

Low

HighSalt solution

Spindle-knot Joint

Open Closed

Time

Salt Low surface energy

High surface energy(nanoporous structure)

Water

CaCl2

Flow

CaCl2

Grooved channel Sample

Grooved spinning duct

Sample

Groove

Neuron

Neurite

Smooth Grooves

Length of spindle-knot (mm)

Vol

ume

of w

ater

(µl

) r2 = 0.97

h

10

20

30

1.2 1.5 1.8 2.1

JointSpindle-knot

Water droplet

d

Figure 3 | Structural coding of fibres. a, Schematic of a digital control scheme for generating artificial spindle-knots and joints by means of topographymodulation and nanoporous structure coding (Supplementary Movie S3). b, SEM image of a magnified spindle-knot where the rough and porous surface isapparent. c, Schematic of a fibre with a different surface energy and water-collection mechanism. d, Artificially spun fibres with spindle-knots and joints. e,Aggregation of water droplets at the artificial spindle-knots. f, Volume of water as a function of spindle-knot length. Data are expressed as mean±s.d.

(n≥ 10). g, Schematic of groove coding on the fibre surface using a grooved round channel. h, Grooved fibres fabricated using the spinning chip. i,Schematic of neuron alignment on a grooved fibre (left) and fluorescence micrograph of neurons on a fibre without grooves (middle) and on a groovedfibre (right) (green, neurofilament; blue, nucleus). Scale bars, 200 nm (b), 1 mm (d), 20 µm (h) and 50 µm (i).

compositions to be coded along the length of the fibre. As shownin the control scheme of Fig. 4a, the periodic modulation of fibrediameter and the coding of varying composition were carriedout simultaneously. Figure 4b,c shows fluorescence micrographs ofthe fibres, indicating that fibres with spatiotemporal variations inmorphology and chemical composition could be produced usinga single microfluidic platform and digital control scheme. Thediameter of the fibre could be periodically tapered by controllingvalve operation. Figure 4d shows the control scheme to producetapered fibres. By periodically opening one and two valves, taperedfibres were produced (Fig. 4e) and multiple 3 µm grooves wereengraved on the fibre surface, as shown in Fig. 4f. In addition toaqueous gels, gas could also be coded along the fibres. This wasachieved by injecting gas and detergent into the alginate solutionchannel to fabricate microfibres that contained gas bubbles. Byoperating the valve at controlled intervals, gas could be distributedeither periodically or uniformly throughout the fibres (Fig. 4g).Figure 4h shows an image of a gas-coded fibre in which the periodand the size of the gas bubbles were controlled by changingthe injecting gas pressure. Figure 4i shows an scanning electronmicroscopy (SEM) image of a gas-coded fibre. Although gels have

a relatively high porosity (the diffusion coefficient of oxygen inalginate is 3.66 × 10−10 m2 s−1; ref. 26), after implantation thepresence of a hypoxicmicroenvironment results in cell death beforethe formation of blood vessels. Thus, gas-coded fibres could beof potential benefit for providing a temporary source of oxygento encapsulated cells.

We further demonstrated the feasibility of fibres for tissueengineering by generating three-dimensional structures withcontrolled cellular organization. As shown in Fig. 5a, weencapsulated primary rat hepatocytes and L929 fibroblasts alongthe length of the fibre either individually or as co-cultures (seeMethods for details). Figure 5a, inset, shows an optical imageof a fibre consisting of three regions and the magnified imagesof each region. In the co-cultured region, the fibroblasts andhepatocytes were coded in parallel (Fig. 5b and SupplementaryFig. S8). Figure 5c illustrates that most of the cells remained viableafter being cultured for 1 and 3 days either individually or inco-culture. Interestingly, the fraction of viable cells in hepatocyte-only cultures decreased throughout the 5-day experiment, whereasthe cells in co-cultures better maintained their viability. Thisresult indicates that the spatiotemporal control of co-culturing





NATURE MATERIALS DOI: 10.1038/NMAT3108 LETTERS

R

G

B

a cb

df

g h i

e f

Opening of two valves

Time

ClosedOpen

SA

SB

Time

Open Closed

Air

SA

Time

ClosedOpen

3 µmJoint

Air drop

Figure 4 | Hybrid coding of fibres. a, Schematic of a digital control scheme to generate a fibre with spatiotemporal variations in morphology and chemicalcomposition (R, alginate solution with red fluorescent polystyrene beads; G, green; B, blue). b,c, Fluorescence micrograph (b) and magnification (c; showingthe area outlined in orange in b) of an embossed fibre coded serially with polystyrene beads (Supplementary Movie S8). d, Schematic of a digital controlscheme for generating tapered fibres (SA, alginate solution A; SB, alginate solution B). e, Micrograph image of a tapered fibre (Supplementary Movie S4).f, The grooved surface of a tapered fibre (orange outline in e). g, Schematic of a digital control scheme for coding gas bubbles in a fibre. h,i, Light microscopy(h) and SEM (i) images of a fibre coded with air bubbles (Supplementary Movie S5). Scale bars, 5 µm (f), 50 µm (i), 1 mm (b,h) and 300 µm (c,e).

within the scaffold is of potential benefit in tissue-engineeringapplications and cell-biology studies. To further demonstratethe use of chemical coding in the study of cell behaviour, wecoded a neutrophil chemoattractant, formyl-Met-Leu-Phe (fMLP,concentration: 10 µM), periodically in a fibre (Fig. 5d) and placedthe fibre inside a culture containing neutrophils (see Methodsfor detail). The presence of fMLP in neutrophil cultures inducedimmediate cell polarization and subsequent migration towardsthe fMLP source. Figure 5e shows fluorescence micrographs atthe boundary between two segments of the fibre after 30min, 3and 6 h of exposure. As can be seen, the neutrophils migratedtowards the fMLP-coded region, as tracked by the arrows. Thisbehaviour is quantified in Fig. 5f. Thus, the coded fibres may beused to modulate the migration of surrounding cells in a spatiallycontrollable manner.

We have presented a spinning method that uses a microfluidicchip combined with a digital fluid controller for the versatile tuningof chemical and morphological properties on the micrometrescale. The coding of diverse materials on the microscale mayoffer new opportunities to generate a range of useful functions inmaterials. Nature also uses spatial patterning of existing materialsto generate new functionality. The proposed method can be usedto fabricate many other materials in a spatially regulated mannerto generate functional fibres of benefit for a variety of applicationsincluding tissue engineering and drug delivery. Furthermore, thismicroengineering approach may encourage innovation in a widerange of application areas in the future, such as water purification,neuron-cell alignment for nerve regeneration and development ofnew functional textiles.

MethodsValve control system. To operate the small pneumatic valves, compression andvacuum pumps were employed. Supplementary Fig. S1 illustrates the set-upfor valve control. Continuous air pressure (around 200 kpa) or vacuum wassupplied, and the ‘on–off’ switching of air flow was controlled using solenoidvalves. The ‘open’ and ‘closed’ operations of the solenoid valve were controlledusing a program written in Lab View (NI Lab View 8.6.1, National Instruments).Communication between the Lab View routine and the solenoid (ST2001-15DN,DKC) proceeded through a data acquisition board (NI DAQ-PAD 6015/6016,National Instruments).

Flow focusing and winding system. The end of the chip consisted of cylindricalchannels with various diameters to generate the coaxial flow, and the sheathflow plays a key role to focus the sample flow stream (Supplementary Fig. S4a).Supplementary Fig. S4b shows an SEM image of a coaxial channel. SupplementaryFig. S4c illustrates the winding system for extruded fibre and the wound fibreson the spool (Supplementary Fig. S4c, inset). By altering the motor speed, thefibres were successfully wound around the rotating rectangular spool (acrylspool: width 15mm×20mm, height 2mm) without breaking. A d.c. gearedmicromotor (PGM30-2838E, Motor Bank, Korea) was used in the winding systemand the range of d.c. used was from 6V to 12V. The relationship between theflow rates and the maximum winding velocity of the motor is illustrated inSupplementary Fig. S4d.

Co-culture of primary rat hepatocytes and fibroblasts in the fibres. Hepatocytesand fibroblasts were co-cultured in the fibre using primary rat hepatocytes andL929 (mouse fibroblast cell line) cells. L929 cells were cultured in Dulbecco’smodified Eagle’s medium (DMEM, Gibco) supplemented with 10% fetal bovineserum, 100 µg streptomycin and 100 unitml−1 of penicillin. Cells were embeddedin the fibre using a spinning chip with three sample channels. First, we dispersedhepatocytes and L929 cells in a 1wt% alginate solutionmixed with a 2wt% chitosansolution and a 1wt% alginate solution at a concentration of 5×106 cellsml−1.The solution with hepatocyte cells was introduced through the middle channelat an average flow rate of ∼10 µl min−1. The solution with the L929 cells wasintroduced through the side channels at the same flow rate. Ten minutes afterfibre generation, the fibre was moved to the cell-culture well containing primary






A

BCD E

*****

ba

c

d

f

A B C D E

e

Parallel cell coding Hepatocytes alone

Fibroblasts alone

fMLPfMLP

30 min 3 h 6 h

Neutrophil

fMLP

Time

Neu

trop

hil n

umbe

r

Days

Perc

enta

ge o

f cel

l via

bilit

y (%

)

HepatocyteHepatocyte /fibroblast

250

200

150

100

50

30 min 3 h 6 h

100

80

60

40

20

1 3 5

Fibroblast

Hepatocyte

Neutrophil

Figure 5 | Spatially coded microfibres for three-dimensional tissue culture. a, Schematic of a periodically coded fibre with either primary rat hepatocytes,fibroblasts (L929) or a mixture of hepatocytes and fibroblasts. Insets, optical image of cells in a fibre (bottom right) and micrographs of embedded cells ina fibre (fibroblasts alone, bottom left; mixture of fibroblasts and hepatocytes, top left; hepatocytes alone, top right). b, Magnified image of a co-cultureregion that consists of multiple parallel layers of either hepatocytes or fibroblasts. c, Hepatocyte and L929 cell viability in two different fibres during a5-day experiment. Data are expressed as mean±s.d. (n≥ 10; ∗, P< 0.005; ∗∗, P< 0.0005; two-tailed t-test). d, Schematic of a fibre periodically codedwith fMLP, which mediates neutrophil migration; inset, fluorescence micrograph of the boundary between two segments of a fibre (green, neutrophil; red,fMLP). e, Time-lapse images at the boundary between two segments of a fibre at 30 min, 3 h and 6 h (the arrows represent time-lapse traces of 3 cells).f, The cell number of each region (A to E) over time. Scale bars, 1,000 µm (a) and 100 µm (b,e).

hepatocyte culture medium. The cells were incubated in an incubator over the next5 days. Cell viability was measured using a live/dead assay (L3224, Invitrogen) eachday, following the protocol mentioned above.

Neural-cell culture on fibres. Grooved fibres and non-grooved fibres wereprepared as previously described. After winding fibres on a spool, the fibreswere dehydrated using ethanol ranging from 50% to 100%, and then driedin an oven at 80 ◦C overnight. Dried fibres were coated with poly-l-ornithinehydrobromide (0.1% in deionized water, Sigma) and laminin (5 ugml−1 indeionized water, Roche) for 10min and 1 h respectively. Neuron cells isolatedfrom rat embryos were diluted to 1×106 cellsml−1 and then seeded on fibres.Cell-culture medium was changed every 2 days. After 5 days, the neuronal cellswere immunostained to identify neural alignments on fibres. Fibres with cellswere fixed in 4% paraformaldehyde for 4 h at 4 ◦C and permeabilized by 0.1%Triton X-100 for 20min at room temperature. Next, fibres with cells wereblocked with 3% BSA for 30min and incubated with primary antibody at 4 ◦Covernight. Primary anti-GFAP antibodies (StemCell Technologies), which werespecific for neurofilament, were used to visualize the location of the neuron.After incubating samples overnight, fibres with cells were washed with 0.1% BSAfor 5min. Alexa Fluor 488 goat anti-rabbit secondary antibody (1:500 dilution,Invitrogen) was applied for 1.5 h at room temperature and washed with 0.1% BSA.Fluorescence micrographs were acquired using confocal microscopy (Olympus)after counterstaining with 4,6-diamidino-2-phenylindole dihydrochloride (DAPI,Invitrogen). All reagents were diluted with pure neurobasal medium to prevent thealginate fibre from dissolving.

Neutrophil migration. An alginate solution of 1% concentration was mixedwith 10 µM fMLP (Invitrogen), 0.1% dimethylsulphoxide and 0.05% fluorescentnanobeads. The mixed solution was then periodically coded into fibre. After

winding the coded fibre on a spool, it was preserved at room temperature (driedfibre is stored at 4 ◦C). Fully differentiated neutrophils were diluted to a densityof 5×105 cellsml−1 in culture medium and seeded on a culture chip. After cellinoculation, fMLP-coded fibres were placed on a culture channel, and covered bya glass slide. To observe neutrophil migration, time-lapse images were acquiredusing a fluorescence microscope (AxioVision 4) for a time duration of 8 h(Supplementary Movie S10).

Received 3 February 2011; accepted 1 August 2011; published online4 September 2011

References1. Daher, R. J., Chahine, N. O., Greenberg, A. S., Sgaglione, N. A. & Grande, D. A.

New methods to diagnose and treat cartilage degeneration. Nature Rev.Rheumatol. 5, 599–607 (2009).

2. Silva, G. A. et al. Selective differentiation of neural progenitor cells byhigh-epitope density nanofibers. Science 303, 1352–1355 (2004).

3. Moutos, F. T., Freed, L. E. & Guilak, F. A biomimetic three-dimensional wovencomposite scaffold for functional tissue engineering of cartilage. Nature Mater.6, 162–167 (2007).

4. Shi, J. J., Votruba, A. R., Farokhzad, O. C. & Langer, R. Nanotechnology indrug delivery and tissue engineering: From discovery to applications.Nano Lett.10, 3223–3230 (2010).

5. Greiner, A. & Wendorff, J. H. Electrospinning: A fascinating methodfor the preparation of ultrathin fibres. Angew. Chem. Int. Ed. 46,5670–5703 (2007).

6. Sill, T. J. & von Recum, H. A. Electro spinning: Applications in drug deliveryand tissue engineering. Biomaterials 29, 1989–2006 (2008).





NATURE MATERIALS DOI: 10.1038/NMAT3108 LETTERS7. Hu, M. et al. Hydrodynamic spinning of hydrogel fibers. Biomaterials 31,

863–869 (2010).8. Rammensee, S., Slotta, U., Scheibel, T. & Bausch, A. R. Assembly mechanism

of recombinant spider silk proteins. Proc. Natl Acad. Sci. USA 105,6590–6595 (2008).

9. Pregibon, D. C., Toner, M. & Doyle, P. S. Multifunctional encoded particlesfor high-throughput biomolecule analysis. Science 315, 1393–1396 (2007).

10. Xu, Q. B. et al. Preparation of monodisperse biodegradable polymermicroparticles using a microfluidic flow-focusing device for controlled drugdelivery. Small 5, 1575–1581 (2009).

11. Breslauer, D. N., Muller, S. J. & Lee, L. P. Generation of monodispersesilk microspheres prepared with microfluidics. Biomacromolecules 11,643–647 (2010).

12. Chung, S. E., Park, W., Shin, S., Lee, S. A. & Kwon, S. Guided and fluidicself-assembly of microstructures using railed microfluidic channels. NatureMater. 7, 581–587 (2008).

13. Mandal, S., Bhaskar, S. & Lahann, J. Micropatterned fiber scaffolds for spatiallycontrolled cell adhesion.Macromol. Rapid Commun. 30, 1638–1644 (2009).

14. Bhaskar, S., Hitt, J., Chang, S. W. L. & Lahann, J. Multicompartmentalmicrocylinders. Angew. Chem. Int. Ed. 48, 4589–4593 (2009).

15. Jeong, W. et al. Hydrodynamic microfabrication via ‘on the fly’photopolymerization of microscale fibers and tubes. Lab Chip 4,576–580 (2004).

16. Breslauer, D. N., Lee, L. P. & Muller, S. J. Simulation of flow in the silk gland.Biomacromolecules 10, 49–57 (2009).

17. Kang, E., Shin, S. J., Lee, K. H. & Lee, S. H. Novel PDMS cylindrical channelsthat generate coaxial flow, and application to fabrication of microfibers andparticles. Lab Chip 10, 1856–1861 (2010).

18. Vollrath, F. Biology of spider silk. Int. J. Biol. Macromol. 24, 81–88 (1999).19. Vollrath, F. & Knight, D. P. Liquid crystalline spinning of spider silk. Nature

410, 541–548 (2001).20. Shin, S. et al. ‘‘On the fly’’ continuous generation of alginate fibers using a

microfluidic device. Langmuir 23, 9104–9108 (2007).21. Kang, E., Lee, D. H., Kim, C. B., Yoo, S. J. & Lee, S. H. A hemispherical

microfluidic channel for the trapping and passive dissipation of microbubbles.J. Micromech. Microeng. 20, 045009 (2010).

22. Hwang, C. M., Khademhosseini, A., Park, Y., Sun, K. & Lee, S. H. Microfluidicchip-based fabrication of PLGA microfiber scaffolds for tissue engineering.Langmuir 24, 6845–6851 (2008).

23. Jung, J. H., Choi, C. H., Chung, S., Chung, Y. M. & Lee, C. S. Microfluidicsynthesis of a cell adhesive Janus polyurethane microfiber. Lab Chip 9,2596–2602 (2009).

24. Zheng, Y. M. et al. Directional water collection on wetted spider silk. Nature463, 640–643 (2010).

25. Blackledge, T. A. & Hayashi, C. Y. Silken toolkits: Biomechanics of silk fibersspun by the orb web spider Argiope argentata (Fabricius 1775). J. Exp. Biol. 209,2452–2461 (2006).

26. Renneberg, R., Sonomoto, K., Katoh, S. & Tanaka, A. Oxygen diffusivityof synthetic gels derived from prepolymers. Appl. Microbiol. Biot. 28,1–7 (1988).

AcknowledgementsThis study was supported by a grant from the NRL (National Research Laboratory)programme, the Korea Science and Engineering Foundation (KOSEF), Republic of Korea(No. 20110020455), basic Science Research Program through the National ResearchFoundation of Korea (NRF) funded by theMinistry of Education, Science andTechnology(No. R11-2008-044-02002-0) and the Korea Research and Engineering Foundation(KOSEF) funded by the Korea government (KRF;No. KRF-2008-220-D00133).

Author contributionsE.K. designed and carried out the experiments and prepared most of the data; G.S.J.carried out an analytical study of the valve system and assisted with the experiments;Y.Y.C. carried out the cell-based evaluation; K.H.L. optimized the material and developedthe flow analysis; A.K. consulted on the manuscript and contributed in writing the paper;S-H.L. proposed the idea,managed the research process andwrote the paper.

Additional informationThe authors declare no competing financial interests. Supplementary informationaccompanies this paper on www.nature.com/naturematerials. Reprints and permissionsinformation is available online at http://www.nature.com/reprints. Correspondence andrequests for materials should be addressed to S-H.L.





http://www.nature.com/reprints


digitally tunable physicochemical coding of material ...tissueeng.net/lab/papers/nmat3108.pdf ·...

Documents